The process - deep dive into Elixir processes

Not wanting to hijack the post, but if I am not in mistake @josevalim said once that most probably we wouldn’t have an Elixir V2.0, but after all it seems to be planned one???

There are a bunch of those comments in elixir core, but I guess not because it’s planed, but because it’s useful if it eventually becomes an option.

Huh, right! Thank you for clarifying! Tried on simple example it worked as expected!

In the project I’m working on turned out the issue was in a different module where async processing is implemented via GenServer handle_info.. I misread the error message

Thank you for bringing it up here!
I just replaced stateless GenServers with supervised tasks and tested with Mox with async: true (yay!)

Each time someone changes a linear-flow GenServer to a Task, an angel gets their wings lol.

I heard that some part of the Phoenix framework is some processes that are not GenServers but still can be supervised by a supervisor. I wonder what’s the minimal requirement to make a process that can be supervised.

There are many non GenServer processes, which can be supervised: Task, :gen_statem and some other :gen_* behaviours. Basically there are plain processes, which are just barebone, then there are processes spawned by :proc_lib, which also adhere to OTP design principles and then there are higher level abstractions and behaviours, which often use :proc_lib internally. Those last ones are the tasks, genservers, statemachines, ….

Then how many messages should a process be able to handle so that it can be supervised? None?

Also, what messages should a supervised process be able to spit out? On what condition?

Those things are afaik documented in the docs for :proc_lib/:sys, like here:

sys and proc_lib — Erlang System Documentation v29.0.2

Not sure if there’s more, I never did a deep dive into the topic.

Thanks a lot for the documentation. That’s fascinating. I’ll give it a shot (in Elixir, of course) this weekend.

I’m pretty sure the only requirement is that you be able to launch the child process with some MFA that returns one of {:ok, pid}, {:error, reason}, or :ignore, where pid is linked.

Tasks don’t emit messages, so that’s not a requirement. Nor do they accept messages, except for brutal_kill messages (that’s built into the VM, so no need to write any code for it). You can choose to handle a soft kill message, but that is optional, because a supervisor knows to brutal_kill after the grace period.

Also note that DynamicSupervisor is elixir only.